MIPS now part of AI startup Wave Computing

June 14, 2018 // By Peter Clarke
Wave Computing Inc. (Campbell, Calif.) has acquired processor IP licensor MIPS Technologies Inc. (Santa Clara, Calif.) according to reports.

Wave Computing is a private company with a history in asynchronous logic that dates back to 2008 but it has more recently opted to be a provider of deep learning systems based on silicon that uses a dataflow architecture to exploit data and model parallelism in convolutional and recurrent neural networks.

MIPS is even more venerable with a history dating back to 1984 and a major installed base of licensees that use its 32 and 64bit RISC processor cores. But the company ended up in the hands of Imagination Technologies and was sold off to Tallwood Ventures in 2017 as part of the sale of Imagination to Chinese interests (see Imagination, MIPS to be sold to China-, California-connected VCs).

By acquiring MIPS, Wave will gain access to a much broader set of customers in embedded applications and with the option to produce hybrid systems that combine conventional software-programmable processing with neural network acceleration for energy efficiency.

It is not known how much Wave Computing paid for MIPS but Tallwood Venture Capital (Palo Alto, Calif.) agreed to paid $65 million for the processor licensor in 2017.

The deal is a fit not least in that a number of Wave Computing executives are former MIPS employees, including CEO Derek Meyer, who was previously vice president of sales and marketing there.

Dado Banatao, is managing partner at Tallwood Venture Capital,which is an investor in both MIPS and Wave Computing and prior to the acquisition was on the board of both.

And in March of 2018 Wave Computing announced it had selected a 64bit multithreaded processor core from MIP for inclusion in future artificial intelligence systems.

Next: Widening the business model


s

Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.